William Check, PhD
To describe the central role that molecular laboratories in community hospitals played in this spring’s pandemic of swine-origin influenza A virus (S-OIV), we might paraphrase T.S. Eliot’s poem “The Hollow Men”:
Between the patient
And the pandemic
Falls the pathologist.
Between the illness
And the incidence
Falls the laboratory.
Members of the lay public encounter a virus pandemic at two main points: when a sick person goes to the doctor’s office or the emergency department and when a person learns from the news that a novel virus strain is circulating widely. In the first instance, laboratories help to diagnose the patient’s illness and identify the pathogen. In the second, they provide the data that state public health departments aggregate to obtain local incidence estimates and that are sent to the Centers for Disease Control and Prevention, where national figures are derived.
These activities occur in any flu season. In the S-OIV pandemic that began in late April, when a novel, swine-derived influenza A virus appeared that had unknown pathologic properties and was not differentiated by existing serologic or molecular assays, laboratories took on a new responsibility—rapidly devising and validating assays for the virus. At that time there were no approved tests for S-OIV. A PCR assay quickly developed by the CDC and approved by the FDA under Emergency Use Authorization was distributed to public health labs. On July 27, also under EUA, the FDA approved a commercial assay for certified labs.
With no commercial assay for S-OIV, community hospital laboratories could continue to perform rapid assays, which identify about half of patients infected with influenza A but do not show whether it is the swine strain. In the face of mounting public concern, an incomplete answer like that was unacceptable. For an adequate answer, samples positive for influenza A could be sent to public health laboratories to determine whether they contained a seasonal strain or S-OIV (sometimes called 2009 influenza A [H1N1]). However, if all hospital laboratories had taken this approach, public health laboratories would have been overwhelmed. So it was particularly fortunate that qualified molecular pathologists in many community hospitals embraced the challenge and either generated new molecular assays to identify S-OIV or validated commercial assays that were not approved for S-OIV but that allowed probable identification of the virus by inference.
One veteran molecular diagnostic laboratory in a high-incidence urban area already had a test up and running that definitively diagnosed S-OIV, though that became apparent only when the lab got its first case of the swine virus. “In fall of 2008, when CDC announced that H1 flu was 99 percent resistant to Tamiflu, our infectious disease physicians asked us if we could set up an assay to differentiate H1 from H3, for purposes of antiviral management. We did so, and fortunately found ourselves with a highly sensitive and specific assay,” explains Karen L. Kaul, MD, PhD, board of directors chair of molecular pathology and director of the Molecular Pathology Division, NorthShore University HealthSystem (NSUHS), and clinical professor of pathology, University of Chicago Pritzker School of Medicine. “We found a publication describing a real-time RT-PCR assay that discriminates H1 from H3 by melt curve analysis” (Stone B, et al. J Virol Methods. 2004;117:103–112).
Responsibility for setting up the real-time assay fell to Kathy Mangold, PhD, a research scientist at NSUHS and clinical assistant professor at the University of Chicago. “We were fortunate that when our clinicians asked for an assay to distinguish the two seasonal flu strains, there was already a publication saying how to do that,” Dr. Mangold says. In this assay, which amplifies part of the viral matrix protein gene, melt curves for human H1 and H3 separate by five degrees. The laboratory began offering the assay for clinical use in February of this year.
“During the first few months that we did the assay, we saw rare variants that melted differently than the seasonal H1 and H3 types, which we called ‘atypical,’” Dr. Mangold says. On April 22 they received a specimen that melted uniquely, right between H1 and H3. “We reported it as ‘unable to subtype,’” Dr. Kaul says. It was that weekend that what Dr. Kaul calls “the media blitz” about swine flu started. “I was thinking, This is what we’ve got,” she says. On April 29, they sent the initial specimen, along with five others with similar melt peaks, to the Illinois Department of Public Health (IDPH) as possible S-OIV specimens. Using the CDC assay, the IDPH laboratory confirmed the presumptive identification, signaling recognition of the first cases of S-OIV in the Midwest. “When IDPH confirmed it as S-OIV, that showed that our assay directly discerned all three subtypes,” Dr. Mangold says, referring to human H1 and H3 and swine H1.
Simultaneously, a second focus of S-OIV in the U.S. arose in the New York City borough of Queens. “We were thrown into it without any choice,” says Christine C. Ginocchio, PhD, director of microbiology, virology, and molecular diagnostics at North Shore-Long Island Jewish Health System Laboratories. “Our hospitals were at the epicenter of the New York City outbreak.” On the weekend of April 24, approximately 75 students from St. Francis Preparatory School showed up in the emergency departments of several hospitals of the North Shore-LIJ Health System. Several had just come back from spending spring break in Mexico. In addition, in the week after returning, “Infected students had passed influenza S-OIV very nicely to other students,” Dr. Ginocchio says. “In just a few days our labs were overrun with influenza testing” (MMWR. 2009;58:470–472).
At the time, seasonal flu was winding down, to about 125 specimens per day. Within a few days, however, the laboratory was receiving 800 samples for flu testing daily, presumably part of the emergent swine flu epidemic. “We quickly realized that the influenza A–B rapid antigen test wasn’t going to do the job,” Dr. Ginocchio says. “It is too insensitive and not able to differentiate seasonal H1 and H3 flu strains from the swine flu variant.” Her laboratory also performs direct fluorescent antibody (DFA) and viral culture on rapidantigen– negative samples, but neither of those assays differentiates seasonal flu strains from S-OIV.
Dr. Ginocchio quickly decided to convert all flu testing to multiplex PCR using the Luminex xTAG Respiratory Virus Panel (RVP), which they were already using in a large research study. RVP can detect and differentiate seasonal influenza A H1 and H3 strains. Though the system is FDA-cleared for many respiratory viruses, for S-OIV it is classified as a lab-developed test.
“When we first started using RVP, we did not know if it accurately detected S-OIV,” Dr. Ginocchio says. “That first weekend we identified 100 possible S-OIV cases (influenza A-positive but unsubtypable as either H1 or H3) with RVP and sent them to the New York State Department of Health lab, which had the CDC assay.” All were confirmed as S-OIV.
What was initially called swine flu and subsequently named S-OIV was first recognized in two children in California in late March. Both children were enrolled in studies that required strain subtyping. When local laboratories could not subtype the influenza A strains, the specimens were sent to the CDC, where sequence analysis determined the virus was swine influenza A (H1N1). In mid-April Mexico informed the World Health Organization of unusually high rates of flu-like illness, in two Mexican states, that was spreading by human contact.
Genetic analysis showed that the cases in California and Mexico were due to the same influenza A strain, which was derived from a genetically complex swine influenza A (H1N2) virus known since 1998 to inhabit North American pigs. This strain, which rarely infected humans, had newly acquired two gene segments from the Eurasian swine lineage—one coding for N1 neuraminidase and the other for matrix protein (see figure in: Morens DM, et al. N Engl J Med. 2009;361:227).
When the sequence of the S-OIV matrix gene was published, Dr. Mangold at NSUHS used GenBank and other publicly available sequence software to determine that their probe should hybridize with it and that the mismatches would theoretically result in a melt peak intermediate between the peaks for seasonal H3 and H1—exactly what they saw. This example “emphasizes how a skilled molecular lab can quickly adapt assays to a new virus because we have building blocks and can make new probes in just a few days,” she points out.
In practice, samples submitted to the NSUHS molecular diagnostics laboratory as suspected influenza were first run on the Eragen multiplex assay. Those that were positive for influenza A were then run on the real-time RT-PCR subtyping assay. Samples that were A-positive with a melt curve between those for H1 or H3 were called “presumptive S-OIV.” Using the EasyMag (BioMérieux) for automated nucleicacid preparation and the LightCycler for real-time PCR, the laboratory was able to run up to 10 batches of 24 samples each per day. Turn-around time to a definitive result was 24 hours.
“At our maximum, we were running 200 samples per day,” Dr. Kaul says. “During regular flu season we run about 2,000 tests and perhaps about 25 percent are positive.” During the S-OIV pandemic, they ran more than twice that number in about 10 weeks, with positivity rates up to 50 percent during some weeks in May.
To validate the assay rigorously, Dr. Kaul says, “we did what you do with any lab-developed test”—they sent about 400 “presumptive S-OIV” samples to the Illinois Department of Public Health. All were verified. “They essentially validated our assay,” says Jan A. Nowak, MD, PhD, director of the Molecular Diagnostics Laboratory at NSUHS.
When Dr. Nowak surveyed other major hospital laboratories in the Chicago area, he found that three had adopted strategies to identify S-OIV. Two were doing RVP and one was using Prodesse and sending influenza A-positive samples to the state. By week four, Dr. Nowak says, “most tests [for suspected S-OIV] were being done in community hospital laboratories.” He estimates that NSUHS and ACL Laboratories, also in Chicago, performed almost 10,000 tests during the pandemic and that the four community hospital laboratories together reduced the state laboratory’s workload by more than half.
Knowing that S-OIV infection was increasing in the community the hospital serves was useful for infection control, says Ari Robicsek, MD, medical director of infection control and hospital epidemiologist at NSUHS and clinical assistant professor of medicine at the University of Chicago Pritzker School of Medicine. He put out several messages to clinicians each week. “At first I simply told doctors to be suspicious and to test for swine flu in persons with a febrile respiratory illness who had visited Mexico, California, or Texas,” he says. As it became clear that the virus was spreading locally, he dropped the geographic clause.
Because the virus didn’t appear to cause death or severe disease, Dr. Robicsek endorsed the CDC directive to test only patients with severe respiratory disease or those with family members at risk of developing severe respiratory disease if they became infected, such as children with severe asthma. If patients with high-risk family members were positive for S-OIV, he recommended giving the at-risk person prophylactic antivirals. “We could do this because we had an accurate high-volume assay,” he says.
Having an accurate assay for S-OIV was also “very helpful” in allowing hospital employees to return to work before seven days if they felt well, Dr. Robicsek says. “If they were negative on real-time PCR, we assumed they were not shedding.” In general, Dr. Kaul says, “For employers asking for return-to-work swabs, I feel comfortable with a negative PCR on a properly collected nasopharyngeal swab that virus is not there.” An unexpected benefit of more testing with the real-time assay was a reduction in the use of antibacterials for influenza-like illness, which Dr. Robicsek attributes to “high awareness” of the virus.
While the real-time RT-PCR assay adopted at NSUHS identifies S-OIV directly, the Luminex xTAG RVP identifies S-OIV by inference. The RVP assay has beads for three targets present in influenza A virus—the gene for matrix protein, which is present in all strains, and the genes for H1 and H3 hemagglutinins. “If a specimen is positive for the matrix gene, but not for H1 or H3, that indicates a possible variant, not seasonal flu A,” Dr. Ginocchio explains. “We showed that during this outbreak if the assay detects influenza A that is unsubtypable, there is a very high probability it is S-OIV” (Ginocchio CG, St. George K. J Clin Microbiol. 2009;47:2347–2348). Since this initial report, the New York State Clinical Laboratory Evaluation Program certified the lab to do confirmatory S-OIV testing using the CDC protocol. Additional testing of hundreds of samples called unsubtypable by the RVP assay has confirmed all to be S-OIV. Dr. Ginocchio and her colleagues also showed that the multiplex PCR assay is superior to rapid antigen test, DFA, and R-Mix culture (Ginocchio CG, et al. J Clin Virol. 2009;45:191–195).
“In a week we converted all our testing to RVP,” Dr. Ginocchio says. “That was a very large undertaking.” For the first eight weeks of the pandemic they were receiving on average 300 to 400 specimens per day. The laboratory was working 12 to 18 hours per day, seven days per week. “This put a very large strain on our staff,” Dr. Ginocchio says. On the peak day they tested more than 900 samples with one or another type of flu test (rapid, RVP, culture, or DFA); peak daily PCR volume was 400. More than 35,000 total flu tests of all types were performed during the nine weeks after April 24. “Things finally quieted down” at the beginning of July, she says.
After the first four weeks of the outbreak, seasonal H1 and H3 flu disappeared. “If we saw flu after that time we assumed it was S-OIV,” Dr. Ginocchio says. “In our outreach population we began to discourage testing and stopped subtyping.” After about six weeks she told pediatricians’ offices that if a child had flu at that time, there was no need to test specifically for swine flu. “They have S-OIV. That is the only type of flu we are seeing now,” she told them.
One important feature of the S-OIV pandemic, Dr. Ginocchio says, is that “the case-fatality ratio was very low, certainly compared to seasonal flu.” Few people were admitted to the hospital; most of those who were admitted had underlying disease or were pregnant.
Also notable was how rapidly S-OIV spread. Officials quickly closed schools, but the horses were already far from the barn. Initial cases in the Queens epicenter were almost all in 13- to 17-year-olds, Dr. Ginocchio says. “Later we started to see younger siblings, five to 11 years old. We had whole families come in; they gave it to each other. This virus has a very high transmission rate.” It was most prevalent in young people. “The older population perhaps had some protective immunity. Younger children and teens had never seen those strains,” Dr. Ginocchio speculates.
She sees one positive note in the high penetration of S-OIV into the young population. “The only good thing is that, if you were between five and 18 years of age and living in Queens, you were probably exposed to S-OIV and may be immune for next season.”
A third laboratorian who responded to the challenge of providing definitive results during the S-OIV pandemic was Marie Louise Landry, MD, professor and vice chair, Department of Laboratory Medicine, and director, Clinical Virology Laboratory, Yale University School of Medicine. “Cytospin DFA is our main screening test for influenza virus,” Dr. Landry says. “We use it for inpatients and outpatients and turn it around in two hours, eight to 18 hours a day, seven days a week.” An advantage of cytospin DFA is that it is multiplex, detecting parainfluenza, influenza A and B, and adenovirus. A disadvantage: “It is very manual and very laborintensive, and during the peak of S-OIV, we were doing up to 130 DFAs a day,” Dr. Landry says. In her laboratory, influenza type A and B PCR is ordinarily performed once a day, Monday to Friday. She has compared a rapid test, cytospin DFA, and PCR (Landry ML, et al. J Clin Virol. 2008; 43: 148– 151). “A rapid test can be done by less skilled people than can DFA or PCR,” Dr. Landry acknowledges. “But sensitivity is much lower, 50 percent or less.”
When the S-OIV pandemic began, the laboratory already had in place the CDC PCR assay for influenza A subtypes H1 and H3 to help guide antiviral therapy, since seasonal H1 virus was resistant to Tamiflu. “We ordinarily confine respiratory virus PCR tests to inpatients and people with underlying risk factors, such as immunosuppression,” Dr. Landry says. “When swine flu became an issue, we put in place a new policy: Any specimen that was DFA-positive for flu A we would automatically subtype, both inpatient and outpatient. And if an outpatient physician requested PCR testing and subtyping, we would do it.” They began doing influenza PCR twice a day, seven days a week.
“All we could say was that the specimen was flu A nonsubtypable,” Dr. Landry continues. “We called these samples ‘presumed or suspicious for swine-origin flu A’ and sent them to the state lab.” When novel S-OIV gene sequences were posted online, Dr. Landry ordered commercial primers and probes and set up a S-OIV PCR assay. “We had basically the same assay as the state lab,” she says. “Because of the public health importance, we asked them if they wanted to confirm our results.” They did, and Dr. Landry sent them several hundred samples. “Ultimately, the state lab said they didn’t need to see our samples anymore,” Dr. Landry says. “They were inundated.”
Soon, Dr. Landry’s laboratory, too, was inundated, particularly because of requests for PCR from outpatient pediatric offices. “Initially we honored those requests,” she says. “But after a while we felt the amount of work was too much and sent a statement to doctors’ offices saying we were not going to be doing PCR on outpatients anymore, unless they were at high risk for influenza complications. At that point more than 98 percent of flu A-positives were of swine origin.” Among samples tested by both DFA and specific PCR, cytospin DFA was 81.1 percent sensitive for adult and pediatric patients combined.
The CDC determined on April 17 that the two children in California had been infected by a swine influenza A virus. That finding was posted online April 21 and reported in the April 24 issue of the MMWR. Immediately an opportunity arose to disseminate this news to an important target audience—the Society for Clinical Virology was meeting that week. Nancy Cornish, MD, director of microbiology at Nebraska Methodist Hospital and Children’s Hospital, Omaha, was one of many clinical virologists who got their first briefing on the S-OIV public health emergency during a late-breaker session at the clinical virology meeting on Wednesday, April 22.
“I had no idea what was going to happen,” Dr. Cornish says. On Sunday, April 26, she checked her e-mail and saw the CDC alert and instructions about how to deal with the novel influenza A virus. She immediately called IC nurses at Methodist and Children’s hospitals to discuss the situation, and she called laboratory personnel to alert them to wear protective clothing. Monday they held a meeting to implement CDC guidelines. Dr. Cornish’s experience illustrates how a pandemic can disrupt the health care system even in low-incidence areas. “Our biggest problem was dealing with the worried well,” she says. Though no cases had been reported in Nebraska yet, “People panicked. We had lines out the door of the ED.
“We ran out of rapid testing kits and wound up using DFA,” Dr. Cornish recalls. “We stayed with DFA. Rapid kits have not worked very well, even for seasonal flu. We have had a number of false-positive results with rapid kits and lots of false-negatives.
“We did a lot of triage,” Dr. Cornish continues. For instance, they asked CDC-recommended questions about travel and worked hard to limit testing to those who met the guidelines for testing. As a result, they held test volume to 10 to 20 per day. Also, testing was discouraged in clinics because the clinicians collecting the specimens did not have the appropriate protective clothing or negative air flow rooms. “We told physicians, ‘If you think people are ill and meet criteria for testing, send them to the hospital and our respiratory therapists will collect nasopharyngeal aspirates, which are the best specimen for testing and what our public health lab was requesting.’ We didn’t want the people collecting specimens getting exposed and possibly infected.”
The course of the epidemic was very different in Omaha from Chicago or New York. “Half of our cases of S-OIV were reported in [early July],” Dr. Cornish says. And only in early July did the Douglas County Health Department alert providers to community transmission.
In Dr. Cornish’s opinion, the public health system worked well in this pandemic. “Nebraska HHS was great,” she says. “I thought CDC did a great job. The Douglas County Health Department and the public health lab kept in constant contact by e-mail. We needed to because the testing and treatment recommendations kept changing as the pandemic evolved.” Still, she says, “I think it was pretty lucky that this outbreak was not particularly virulent. It was a gentle way to test the system to find out what worked and what we need to improve.”
Dr. Ginocchio, too, says, “Our public health labs did an absolutely fantastic job. However,” she adds, “it was totally overwhelming to them as well. This is why labs like ours—sentinel labs—play such a critical supporting role in public health.” She hopes the CDC recognizes the help they gave them during this crisis.
“We served as the front line for public health labs,” Dr. Ginocchio continues. “We saved them a huge amount of work by ruling out seasonal flu, which is something most labs can’t do. If we had not been able to do that, all those samples would have had to go to the department of health. We left the public health lab free to deal with samples from places that did not have that capability.”
Dr. Nowak believes that the CDC is unaware of the contribution of laboratories like NSUHS and ACL. He sees a potential mutual benefit. “They have a lot to gain from working with community hospital labs,” he says. “We decrease their workload. And local labs need their support to validate our labdeveloped tests.”
In the opinion of George Dizikes, PhD, HCLD(ABB), CLIA laboratory director at the Illinois Department of Public Health, the CDC gets high marks. “CDC did a good job in quickly developing and validating PCR reagents, then disseminating them to public health labs,” Dr. Dizikes says. “We were the only ones with access to CDC reagents, so the diagnostic task fell to us. This is not usual.”
Dr. Dizikes also recognizes the contribution of community hospital laboratories. “Here in Illinois some advanced labs were able to help us handle specimens,” he says, naming NSUHS and ACL in particular. “They were able to exclude many samples.”
Dr. Dizikes, like others, is grateful that S-OIV was not more virulent and that lab testing went well. “People are using this as an important learning exercise in planning for our next encounter with this virus,” he says.
In New York state, too, the private-public partnership is recognized. “We were very fortunate we had a good relationship with hospital laboratories, such as Dr. Ginocchio’s, which has an excellent understanding of molecular technologies,” says Kirsten St. George, PhD, MAppSc, chief of the Laboratory of Viral Diseases at Wadsworth Center, New York State Department of Health. “As an epidemic or pandemic progresses, the role of these labs changes. In the very early stages, specimens are sent to state labs for confirmation. As things evolve and this becomes more of an established disease, public health labs take on more of a surveillance role and patient diagnosis falls more to hospital labs.”
Dr. St. George calls the S-OIV situation “somewhat complicated” because molecular methods had to be used for the testing, which meant testing was limited to labs that had advanced molecular capabilities and could bring molecular assays on line quickly. “Smaller community hospital labs don’t have facilities or experience or staff to do that,” she says, predicting, “More laboratories will be able to do definitive diagnosis in the second wave of the S-OIV pandemic.”
“The second wave”—the phrase that falls so expectantly from everyone’s lips. When will it begin? Will S-OIV displace seasonal flu strains? Will it be more severe? Will it have acquired Tamiflu resistance? No one knows the answers to these questions. Which is why surveillance is so important. “What I do is mainly try to set up influenza virus surveillance for summer at selected sites around the state to monitor virological characterization,” says Kenneth Soyemi, MD, MPH, assistant medical director in the Illinois Department of Public Health’s Office of Health Protection. About 80 U.S. World Health Organization collaborating laboratories and 70 National Respiratory and Enteric Virus Surveillance System labs throughout the United States participate in virologic surveillance for influenza. “People are concerned about when S-OIV comes back in the second wave in winter. We hope it is not much more virulent than in the past few months,” Dr. Soyemi says. One determinant of virulence could be the PB1 gene. “If it comes back with the PB1-F1 allele, it could cause mass fatalities,” he says.
What will happen to surveillance effectiveness if S-OIV mutates to evade even current molecular assays? “That could happen,” Dr. Ginocchio says. “If there is a significant enough mutation, any PCR assay would miss it.” It’s not likely that the matrix gene would mutate that quickly, in Dr. Ginocchio’s opinion. Still, if it did, “We wouldn’t really have any idea,” she says.
Alternatively, if another variant comes up, current tests wouldn’t differentiate it from S-OIV. This is why Dr. Ginocchio has approval from the New York State Department of Health to run the “CDC assay” on anything that comes up as unsubtypable in the next flu season. They would use the CDC protocol in a lab-developed format. “Our testing algorithm will be that the physician will request RVP,” Dr. Ginocchio says. “If that comes up positive for influenza matrix protein but not seasonal H1 or H3, it would be further typed using the CDC assay for S-OIV.” Samples that come up negative for S-OIV would be forwarded to the public health lab for further testing.
Dr. Kaul agrees that a fundamental genetic change in S-OIV is a concern. She wonders whether it would be possible to develop a regional or nationwide sentinel system to detect genetic changes in key regions in the virus.
Of course, S-OIV is always madly mutating. It is a virus, after all. Dr. Landry has been looking for signs of this on the CDC assay for S-OIV. “With real-time PCR you get a threshold,” she says. “You see the Ct [cycle threshold], which is an indication of how much virus is present.” If a sample is S-OIV, it should have fairly similar Ct values—within three or four cycles—for swine A and for swine H1. “If there is a big difference in Ct values between swine A and swine H1,” Dr. Landry says, “we would be concerned about genetic variability and forward samples to the CDC.” However, labs should be aware that spurious discrepancies can be seen between the two S-OIV PCRs, depending on the RT-PCR kit employed. “We were using AgPath-ID one-step RT-PCR kit, which worked very well for flu A, seasonal H1 and H3 and swine A, but as we found out later, suboptimally for swine H1.” When things became less hectic, they retested the same samples with the SuperScript III Platinum RT-PCR kit, and the discrepancies resolved. “So it was a false alarm,” she says.
Looking back at the first wave, we can conclude that, in this pandemic, where it all came down to molecular testing, community hospital laboratories were especially important. “Community hospital molecular diagnostic laboratories have extensive expertise in developing molecular assays from published information, they can be rapidly mobilized, and they are driven by patient care,” Dr. Nowak asserts. “This was the first pandemic where we had widespread ability to do molecular testing—sequence detection—even more so than with SARS.” Speaking for molecular diagnostic laboratories everywhere, Dr. Nowak declares, “This one is ours.”
William Check is a medical writer in Wilmette, Ill.